You are on page 1of 24

Multidisciplinary Optimization applied to

the Oil&Gas Industry

Rodrigo Ferraz
Nicolas Spogis
ESSS
Agenda

modeFRONTIER
Example: Impeller Optimization using CFD
and modeFrontier
Why using optimization software?
Not straight to obtain optimized results in:
Problems involving large number of parameters;
Constrained problems;
Multi-objective studies
Ex.: Increase power and reduce consumption at the same time.
Multi-disciplinary analyses;

Optimization software also helps in:


Reducing man time to create models
Help understanding the role of the different parameters

Combining CAE tools and optimization eliminates trial and error


technique

3
Defining modeFRONTIER

modeFRONTIER is a multi-objective optimization and design


environment, written to allow easy coupling to almost any
computer aided engineering (CAE) tool, whether commercial or
in-house.
modeFRONTIER Around the World

Icon UK
Local Distributor

EMT-R
Local Distributor

blueCAPE
Local Distributor
CDAJ
Mindware Local Distributor
Local Distributor ESTECO GmbH

SIREHNA
Local Distributor Albert Turtscher
ESTECO NA Local Distributor
North America

Engin Soft Trading


Local Distributor ESTECO s.r.l.
Headquarter/R&D
ESSS
Local Distributor

ESTECO
Local Resellers
CUSTOMERS: over 150 accounts
modeFrontier
modeFrontier: Multi Objective Design Environment:

Process Integration and Optimization Software;


Allows both single and multi-objective optimization;

Allows for Robust Design and Six-Sigma studies;

Easily wrapped around almost any CAE software;

Several post-processing tools;


Works on Unix, Linux and Windows;
The Concept behind modeFRONTIER

The Black Box:


Generates the outputs
accordingly to the inputs

Input Variables:
Entities that define the
design space.

Output Variables:
Measures from the
system
The Black Box

The black box can be:


A set of solvers that models and solves in a
numerical manner the design problem (e.g.
CAD/CAE tools)
A set of experiments that produces some data
Multi-disciplinary Scenario

CFD CAD
(Fluent, CFX, (CATIA, UGS,
1D tools, etc) PROE)

Others FEM
(Ansys,Nastran,Ma
(In-House codes,
dymo, etc)
MATLAB, Excel)
Process Integration and Optimization
Example: Impeller Optimization using
CFX and modeFrontier

Study conducted by:

Nicolas Spogis, PhD.


E-mail: nicolas@esss.com.br

Jos Roberto Nunhez, PhD


E-mail: nunhez@feq.unicamp.br
Study Objectives

Design of a high efficiently hydrofoil:


Pure Axial Flow (minimum Radial Flow)
Promote solid dispersion
Torque Reduction
Saving Costs

Cortesy of Unicamp: Prof. Jos Roberto Nunhez

Through the use of:


Computational Fluid Dynamics
Multi-Objective Design Optimization
Impeller optimization steps

CAD and Mesh CFD Solver


Icem Scripts
Generation
CFX Scripts

Batch Scripts

Impeller

Multidisciplinary Post Process


Optimization
Input Variables
Variable Minimum Value Maximum Value Discrete / Continuum

Impeller diameter 0.4 0.5 Continuum

Root chord 0.1 0.2 Continuum

Tip chord 0.1 0.2 Continuum

Root chord angle 20 degrees 70 degrees Continuum


(related to rotation axis) (related to rotation axis)

Tip chord angle 30 degrees 95 degrees Continuum


(related to rotation axis) (related to rotation axis)

Root profile DAE11, S1223, E387, FX 63-137 Discrete


17 different profiles
Tip profile DAE11, S1223, E387, FX 63-137 Discrete
17 different profiles
Blade Profiles: Low Reynolds Hydrofoil
Objective Functions

Vessel solid concentration


variance
Heavy solid suspension s
2 1 n

Ci C
n 1 i 1

2

Pure Axial Flow


Lack of Radial Flow

Pumping effectiveness Nq
Torque Reduction
Pefect
Np
Saving Costs
Computational requirements

One evaluation takes approximately 5.2 hours of


computing time;

Optimization requires several evaluations to converge to


an optimum;

Strategy: Combine CFD with Response Surface


Methodology (metamodels);
Using metamodels

Initial DOE Filtered designs


(Eliminate unfeasible and using D-Optimal)

Response Surfaces (RSM)

Genetic algorithms use


the RSM information to
find the optimum point
Pareto Frontier
Non Dominated Points

Dominated Points
CFD Results
Solid supension velocity
Original Design:
Average axial Flow
(presence of radial flow)
High Np Torque
High power consumption
Low solid suspension

Optimum Design:
Pure axial Flow (minimum
radial flow)
Low Np Torque
Low power consumption
High solid suspension
Solid Suspension

Experimental results: Cortesy of Unicamp: Prof. Jos Roberto Nunhez


N=290 RPM ->P = 6.09 W N=172 RPM ->P = 6.06 W

N=390 RPM ->P = 10.54 W N=232 RPM ->P = 11.23 W


Solid Suspension

Experimental results: Cortesy of Unicamp (Prof. Jos Roberto Nunhez)


Final remarks

We would like to thank Professor Jos Roberto Nunhez


and Unicamp for providing the laboratory and helping
with planning and execution of the experimental
analyses presented in this work;

This study is part of the work conducted by Nicolas


Spogis at his doctorate at Unicamp University, under
Professors Jos Roberto Nunhez orientation.
END

Obrigado

Rodrigo Ferraz
ferraz@esss.com.br
+55 (48) 3953-0044

You might also like